Musk's Tesla Cortex Supercomputer

Elon Musk has unveiled the first video of Tesla's new Cortex supercomputer. This cluster will be the brain center for developing Tesla's autonomous driving systems and other projects.

Cortex is located at Tesla's Giga Texas plant. At launch, it will include 50,000 NVIDIA H100 GPUs and 20,000 of Tesla's own chips. According to preliminary estimates, the frame captured about 2,000 GPU servers, which is less than 3% of the planned full capacity! Powering and cooling this giant at launch will require 130 megawatts of energy, and by 2026, this figure will grow to an impressive 500 MW.

The main purpose of Cortex is to train the Full Self-Driving System (FSD) for Tesla cars and future robotaxis. In addition, the supercomputer will be used to develop AI for the humanoid robot Optimus, which is planned to begin limited production in 2025.